Putting weather predictions into the model

Top panel: 10th to 90th percentile multi-model range of monthly rainfall totals for 20th Century (grey) and future period (red). Bottom panel: 10th to 90th percentile multi-model range of monthly rainfall anomalies between the future simulation period and the 20th Century simulation period.

Top panel: 10th to 90th percentile multi-model range of monthly rainfall totals for 20th Century (grey) and future period (red). Bottom panel: 10th to 90th percentile multi-model range of monthly rainfall anomalies between the future simulation period and the 20th Century simulation period.

What I think this graphic shows (and I really do need to check with someone who understand this data properly!) is that current predictions for an area quite near to Somie called Koundja suggest that local farmers may well see quite a bit more rain November through to February, and there being less rain June – September.

What strikes me here is that the predictions suggest an end to the dry season, and less rain in when some crops might be thirstiest i.e. the months leading up to harvest.

If the red area above the grey box for July means that we can assume rainfall could be higher than now, then this could also be a problem since very heavy rains can knock over tall plants like maize.

But this is just conjecture. Ecosystems can be complex and sensitive – it is very difficult to know what these changes in weather will actually mean for local farmers i.e. whether new weeds, insects, molds will emerge, and whether these will be detrimental.

In an ideal world the farmers might systematically try out a variety of new crops and see which bring the most beneficial yields (prices and nutritional value). This could be complicated and risky in the sense that if many of the experiments fail there will be a lower overall yield. Farmers could club together to exchange information to lower the risk.

Maybe in the future simulation models will be good enough to explore the multi-dimensional space of possibilities and narrow down the experiments to the most promising few.

Finally, let’s not forget that rainfall is only one of the many weather-related variables that will effect crops. Here’s a rather alarming-looking projection for temperature:

Top panel: 10th to 90th percentile multi-model range of monthly mean daily maximum temperatures for 20th Century (grey) and future period (red). Bottom panel: 10th to 90th percentile multi-model range of monthly mean daily maximum temperature anomalies between the future simulation period and the 20th Century simulation period.

Top panel: 10th to 90th percentile multi-model range of monthly mean daily maximum temperatures for 20th Century (grey) and future period (red). Bottom panel: 10th to 90th percentile multi-model range of monthly mean daily maximum temperature anomalies between the future simulation period and the 20th Century simulation period.

Posted in Cameroon | 2 Comments

2 Responses to “Putting weather predictions into the model”

  1. Kenneth Kahn says:

    A link to your data source would be good.

    The red is 2081 to 2100 — a long time before they have to worry about that.

  2. […] Howard Noble [ Howard Noble ]. “Putting weather predictions into the model.” Modelling4all Project Blog. University of Oxford, IT Services department. Web. Date of access. 12th March 2013. http://blogs.it.ox.ac.uk/modelling4all/2013/03/12/putting-weather-predictions-into-the-model/ […]

Leave a Reply